8,745 Hits in 4.6 sec

Reducing offline evaluation bias of collaborative filtering algorithms [article]

Arnaud De Myttenaere , Bénédicte Le Grand
2015 arXiv   pre-print
This paper presents a new application of a weighted offline evaluation to reduce this bias for collaborative filtering algorithms.  ...  It thus influences the way users interact with the system and, as a consequence, bias the evaluation of the performance of a recommendation algorithm computed using historical data (via offline evaluation  ...  Thus, this method seems to reduce the bias for the very simple class of constant algorithms. In the next part we apply this method to collaborative filtering algorithms.  ... 
arXiv:1506.04135v1 fatcat:osqalsytu5gbrdpt5jbjhhs6ai

Study of a bias in the offline evaluation of a recommendation algorithm [article]

Arnaud De Myttenaere , Bénédicte Le Grand
2015 arXiv   pre-print
This paper describes this bias and discuss the relevance of a weighted offline evaluation to reduce this bias for different classes of recommendation algorithms.  ...  It thus influences the way users interact with the system and, as a consequence, bias the evaluation of the performance of a recommendation algorithm computed using historical data (via offline evaluation  ...  In the next part we discuss the relevance of this procedure to reduce the offline evaluation bias on collaborative filtering algorithms.  ... 
arXiv:1511.01280v1 fatcat:s4j3pggganfibk4ninktcssubu

BanditMF: Multi-Armed Bandit Based Matrix Factorization Recommender System [article]

Shenghao Xu
2021 arXiv   pre-print
BanditMF is designed to address two challenges in the multi-armed bandits algorithm and collaborative filtering: (1) how to solve the cold start problem for collaborative filtering under the condition  ...  For collaborative filtering, the classical method is training the model offline, then perform the online testing, but this approach can no longer handle the dynamic changes in user preferences which is  ...  To compare the performance between the base method and bias method, Mean Square Error (MSE) is used to evaluate these two algorithms.  ... 
arXiv:2106.10898v2 fatcat:upnm5pmva5agvenkcuolwv5muq

Learning to Rank Research Articles: A Case Study of Collaborative Filtering and Learning to Rank in ScienceDirect

Daniel Kershaw, Benjamin Pettit, Maya Hristakeva, Kris Jack
2020 International Workshop on Bibliometric-enhanced Information Retrieval  
We then describe offline and online evaluation, which are essential for productionizing any recommender.  ...  We first introduce itemto-item collaborative filtering (CF), then how these recommendations are rescored with a LtR model.  ...  Online evaluation Our best candidate algorithms from offline evaluation are compared against the current production model via A/B testing.  ... 
dblp:conf/birws/KershawPHJ20 fatcat:lhqmte2j3ba7tfax2dws47jmiy

RecSys for distributed events

Richard Schaller, Morgan Harvey, David Elsweiler
2013 Proceedings of the 36th international ACM SIGIR conference on Research and development in information retrieval - SIGIR '13  
There are often a large number of events on offer and the times in which they can be visited are heavily constrained, therefore the task of choosing events to visit and in which order can be very difficult  ...  Distributed events are collections of events taking place within a small area over the same time period and relating to a single topic.  ...  A modern, state-of-the-art collaborative filtering algorithm [2] .  ... 
doi:10.1145/2484028.2484119 dblp:conf/sigir/SchallerHE13 fatcat:hqmjacuinncedmo2e6udpcdp54

Triggering on hard probes in heavy-ion collisions with CMS

G Roland, the CMS Collaboration
2007 Journal of Physics G: Nuclear and Particle Physics  
The study is based on measurements of the timing performance of the offline algorithms and event-size distributions using full simulations.  ...  Concentrating on two physics channels, dimuons from decays of quarkonia and single jets, we evaluate a possible trigger strategy for Pb+Pb running that relies on event selection solely in the High-Level  ...  All rejection of Pb+Pb collisions will be based on the outcome of HLT trigger algorithms that are identical to the corresponding offline algorithms or optimized versions of the offline algorithms.  ... 
doi:10.1088/0954-3899/34/8/s84 fatcat:aqnn2xfyqzdfretbstp5eyqzjq

Microsoft recommenders

Scott Graham, Jun-Ki Min, Tao Wu
2019 Proceedings of the 13th ACM Conference on Recommender Systems - RecSys '19  
The purpose of this work is to highlight the content of the Microsoft Recommenders repository and show how it can be used to reduce the time involved in developing recommender systems.  ...  The open source repository provides python utilities to simplify common recommender-related data science work as well as example Jupyter notebooks that demonstrate use of the algorithms and tools under  ...  Special thanks to Andreas Argyriou, Miguel González-Fierro, and Le Zhang for their work in the development of the repository.  ... 
doi:10.1145/3298689.3346967 dblp:conf/recsys/GrahamMW19 fatcat:akiqjj2qbvaxphhu6ttamyfpsq

Personalization in E-Grocery: Top-N versus Top-k Rankings [article]

Franziska Scherpinski, Stefan Lessmann
2021 arXiv   pre-print
In e-grocery, shoppers face an overwhelming number of items, the majority of which is irrelevant for the shopper.  ...  Specifically, the proposed RS reduces IO by 29.4% and lowers users' search time by 3.3 seconds per item. The field experiment also reveals a 7% uplift in revenue due to the top-N ranking.  ...  The developers of COUSIN also demonstrate its superiority over other collaborative filtering methods in an offline evaluation (Gan 2015) .  ... 
arXiv:2105.14599v1 fatcat:izmqwdfh2naanfgxbbls7kry7i

Selecting Appropriate Metrics for Evaluation of Recommender Systems

Bhupesh Rawat, Sanjay K. Dwivedi
2019 International Journal of Information Technology and Computer Science  
However, with the availability of several recommender tasks, recommender algorithms, and evaluation metrics, it is often difficult for a researcher to find their best combination.  ...  This paper aims to discuss various evaluation metrics in order to help researchers to select the most appropriate metric which matches a given task and an algorithm so as to provide good quality of recommendations  ...  Offline Experiments The objective of the offline experiment is to select the most appropriate recommendation approaches and filter out the irrelevant approaches, which reduces the list of candidate algorithms  ... 
doi:10.5815/ijitcs.2019.01.02 fatcat:df2ar6nsj5dh7an7bdvlj4wed4

An Educational News Dataset for Recommender Systems [chapter]

Yujie Xing, Itishree Mohallick, Jon Atle Gulla, Özlem Özgöbek, Lemei Zhang
2020 Communications in Computer and Information Science  
We discuss the structure and purpose of the refined dataset in this paper.  ...  AbstractDatasets are an integral part of contemporary research on recommender systems.  ...  This work was carried out as part of the industry-led research project RecTech, project number 245469, supported by the Research Council of Norway's BIA innovation research program.  ... 
doi:10.1007/978-3-030-65965-3_39 fatcat:7e5lmfw4jbc4bgrzolu7iipste


Azarias Reda, Yubin Park, Mitul Tiwari, Christian Posse, Sam Shah
2012 Proceedings of the 21st ACM international conference on Information and knowledge management - CIKM '12  
Metaphor builds on a number of signals and filters that capture several dimensions of relatedness across member search activity.  ...  Third, we introduce a query length model for capturing bias in recommendation click behavior. We also discuss some of the practical concerns in deploying related search recommendations.  ...  Offline Evaluation Offline evaluation provides a scientific and easily repeatable mechanism for establishing performance and tuning parameters.  ... 
doi:10.1145/2396761.2396847 dblp:conf/cikm/RedaPTPS12 fatcat:4nttbjperngi5dj46fjtivsswe

A User Trust-Based Collaborative Filtering Recommendation Algorithm [chapter]

Fuzhi Zhang, Long Bai, Feng Gao
2009 Lecture Notes in Computer Science  
The values of trust among users are adjusted by using the reinforcement learning algorithm. On the basis of this, a user trust-based collaborative filtering recommendation algorithm is proposed.  ...  Experimental results show that the proposed algorithm outperforms the traditional user-based and item-based collaborative filtering algorithm in recommendation accuracy, especially in the face of malicious  ...  Experimental Results To evaluate our algorithm that combined user trust with the collaborative filtering approach.  ... 
doi:10.1007/978-3-642-11145-7_32 fatcat:z5khsnwkpbbotlvasxbjihrbmm

Cold-start Problem in Collaborative Recommender Systems: Efficient Methods Based on Ask-to-rate Technique

Mohammad-Hossein Nadimi-Shahraki, Mozhde Bahadorpour
2014 Journal of Computing and Information Technology  
Therefore, a major challenge of the collaborative filtering approach can be how to make recommendations for a new user, that is called cold-start user problem.  ...  To develop a recommender system, the collaborative filtering is the best known approach, which considers the ratings of users who have similar rating profiles or rating patterns.  ...  Collaborative Filtering Recommender Systems Collaborative filtering recommender systems are one of the biggest sub-domains of information retrieval.  ... 
doi:10.2498/cit.1002223 fatcat:tzrirpnk65bnbbmgaf4vuifubq

Review of Techniques for Recommender Systems

Harbhajan Kaur, Mohita Garag, Amanjot Kaur
2017 International Journal of Advanced Research in Computer Science and Software Engineering  
It is determined that Bagging reduced variance of unstable methods, while boosting methods reduced both the bias and variance of unstable methods but increased the variance for NaiveBayes was very stable  ...  A few algorithms are compared using some evaluation metric instead of absolute benchmarking of algorithms. The experimental settings are suggested to make decisions between algorithms.  ... 
doi:10.23956/ijarcsse/v7i4/0134 fatcat:kwwzlyfscnhinmfi2rsa4nzmee

Separating-Plane Factorization Models

Haolan Chen, Di Niu, Kunfeng Lai, Yu Xu, Masoud Ardakani
2016 Proceedings of the 25th ACM International on Conference on Information and Knowledge Management - CIKM '16  
With extensive offline evaluation in Tencent Data Warehouse (TDW) based on big data, we show that our approach outperforms a wide range of state-of-the-art methods.  ...  Results show that our approach can increase the video click through rate by 23% over implicit-feedback collaborative filtering (IFCF), a scheme implemented in Spark's MLlib. ii  ...  And the biases for each user, video and day of the week, and the global bias are taken into account.  ... 
doi:10.1145/2983323.2983348 dblp:conf/cikm/ChenNLXA16 fatcat:awyfcrix65ctnd6dbs5yg644uq
« Previous Showing results 1 — 15 out of 8,745 results